If optional labs are explained more like data preprocessing and normalization then it would be 5 stars.

Not bad. You definitely need pre requisites beyond python. An understanding of numpy, pandas and matplotlib. Also although these are used in a lot of the code in this course they are never explained. In fact, many parameters are not explained and the authors of this course did their best job to explain things at a very high level of complication. Not much in this course is broken down in simple terms which would have helped a great deal in terms of moving on to other subjects quicker. You are left to decipher everything.I did the course in 5 weeks I scored a 90% on the final and I did the final optional lab and scored 53/53 100%. This course if authored better could be a 3 week course easily. The material is not that complicated once you break it down in a more relatable way which is once again most of the effort in this course. There are also many formulas presented in this course which you dont really need to know. A simple foot note of them would be enough but they are a large focal point for no good reason. The output that you end up creating in the labs is just different averages of the models you create. They dont show you how to output anything cool like actual predictions rather how well the model is performing. If your a tiny bit clever you can sort this out though but its not the focus which makes it way less fun. So to recap (My opinion):This course is not very fun. The material conceptually is really fun but you might never know from this course. The application of the material in this course could be very fun but in this course its not. The explanations in this course could be fun and inspiring but they are instead rote and boring just like the labs. The positive aspect of this course is its kind of accredited you have something to show from your work, a certificate and there is a structure that you have to follow which helps. They claimed there would be projects that would help on a resume... not really at least nothing someone would be impressed by that didn't look super generic.

This was a good course to see how the basic ML models can be used with clear examples in Python. It was a very good sequel to the Stanford as this course didn't go into detail on the algorithms or any depth in to the math behind the scenes. In fact, you could ignore the equations and still do fine. Unfortunately, I didn't feel I learned enough, specifically about how to tune the parameters and improve the results of different algorithms. The final could be accomplished by simply cutting and pasting the work done in the non-graded 'labs' and providing any level of accuracy scores. I would have welcomed more depth on optimization. Also the hardest part of the course was using matplotlib but you didn't even need to understand it to pass. Overall, I'm glad I took this course. It was very helpful in my learning journey.

Actual content is good, but i deducted two stars. One star because the pacing of the course is just too fast. The course could really be split into two courses: one on regression and one on classification/clustering. I deducted the second star because the assignments really need to be clearer, especially the final assignment. It would greatly help the people doing the assignment *and the people grading it* if there were more explicit prompts for where you wanted to see, e.g. jaccard score for the knn model, or if you said, "build a visualization that demonstrates the accuracy of knn models for all k, 0<k<20". Being more explicit about the expectations would make the assignment a better evaluation of the student's understanding.

The Machine Learning with Python course was very challenging. The final assignment, though, seemed to require knowledge not yet learned, which made it rough to complete. Also, although I completed the notebook, all of my cells were not visible to the reviewer even though my settings were such that all cells should have been visible to him/her. I restarted the kernels and re-ran my code a couple times and it was finally visible when I opened the shareable link. That delayed my receipt of an accurate score for a few days. Ugh.

This course has all the fundamentals and in depth learning videos. For me it had many spelling errors in the material that were either annoying or misleading which for a company like IBM not good enough. The guidance around the final capstone project could have been clearer for me, and I had several issues logging into IBM's Watson Studio. I did like the peer graded assessment strategy.

This course was well paced, however, it did not go into sufficient detail when it came to explaining the fundamentals of machine learning. The final assessment does not appropriately justify the knowledge one was meant to have learnt during the course. More time should be spent understanding how the models work and how best to tune their hyperparameters to achieve the best state.

O curso em si é bom, mas como a maioria dos cursos online não temos espaço para dúvidas (e não, o fórum não é suficiente).

A tarefa final é muito mal explicada.

Também senti falta de mais Python durante as aulas, que só cobrem o aspecto teórico. Embora muito bom, ao se deparar com o código, surgem muitas dúvidas.

It is good overall but missing lots of useful ML algorithms such as boosting algorithms, density and hyrachial clustering algoritms. It also didn't mention about overfitting and underfitting cases comparing with training and test scores.

very bad course

no proper instructions or explanations in videos

I have just completed the course and mentioned below are my key pros and cons for this course:

Pros:

1) I loved the theory and different techniques explained in the course.

2) The presentations were very well made and it helped me to gain knowledge as far as ML is concerned.

Cons:

1) This is a pretty outdated course, where there are ALOT of typos and coding errors throughout the labs as the coder has left IBM and is working in some other company for more than a year now. Thats is why no one is there to update the course.

2) The title of the course should be "Machine Learning with Mathematics" rather than "MAchine Learning with Python" because the emphasis of this course is on using mathematics to solve ML related problems and that is why most of the libraries and techniques used in the python files were not defined.

3) This IBM's specialization is of BEGINNER level and the inclusion of an INTERMEDIATE level course which requires you have to have some experience in Data Science and advanced level knowledge of Python is just mind boggling to me. It would have been great if a basic level course of ML would have been developed which emphasized on explaining while using Python libraries would have been much more appropriate for us.

4) Lastly, it has confused me while going through this course that numerous times the lecturer spent major time of the lecture in explaining the advanced mathematics which Pythons libraries can easily do for you, even if he told us that remembering of the mathematics is not need. STILL he explained it. I don't know why he did it again and again.

I learned a lot from this course. However, had I known what I had to go through to learn the knowledge, I would not have taken the course; the process is too painful. Therefore I would not recommend the course to future learners. Read my review and save yourself $39.

1) Too many typos, bugs, inconsistencies throughout the videos and labs. The same mistakes have been brought up by students over and over again on the discussion forum, but have never been fixed.

2) Teaching staff do not pay attention to students asking for help. Sometimes when they do answer the question, they give a very vague or irrelevant answer; and when being pointed out by students that their answer is not helpful, the teaching staff do not bother to reply and address the issue. I feel like the teaching staff never went through the entire course themselves so they do not understand our students' concern and frustration.

3) A lot of Python codes are never explained or commented. This is a beginner level class but they expect you to be able to code proficiently; otherwise you are going to be stuck with one line of unexplained code for a long time...

4) The whole course is like a giant advertisement for IBM Cloud, which is not user-friendly at all.

A bit dissapointed by this course. The main topics were given clear and simple, but there were too few details, saying that all the details are out of scope of the course. But I would prefer to have more information and also more mathematical details (I find the argument that it needs appropriate background strange: if one wants to learn Machine Learning, should already have some basic mathematical background as knowledge of derivatives, integrals, etc).

Another big disappointment was absence of the graded programming assignments, except the final project. Every part of the course had just graded Quiz, but real hand-on scripting in python was given just as non-graded example, and then final assignment basically consisted from the same code.I find this approach quite useless. Also the final assignment had to be done at the IBM Watson website - I guess just for advertisement of IBM services - but this is useless to waste time on registering there, and figuring out how to do things there, if instead could be done inside coursera itself.

And finally, there few some mistakes and typos e.g. in the final assignment, which made everything a bit confusing.

Like many of the courses, the instructions are not in a format that supports incremental learning and focuses on the mechanics for performing an activity rather than an explanation for why and the reason we are doing these things.

The objectives and measures of success for the final exercise is not clearly articulated, causing me to guess as to what the evaluator had wanted us to do. The instructions said to solve for the four types of methods, but left it to the student as to if they wished to generate graphics, etc. If the only objective was to generate the Jaccard score, F1 score, and LogLoss (as appropriate) to complete the activities, then it should have been stated. In addition, the examples presented in the course labs did not have us generating the F1 and Jaccard scores for many of the models.

The lectures were interesting, however, I do not think that the labs are at all aligned with the knowledge learned during the courses, at all. All videos and reading material are about theory and mathematical concepts, which truly are interesting, but there is zero Python training in this course, a soon as you get to a lab, you need to figure it out on your own. As a reference, AWS labs are 100% guided which makes them way easier to learn. Also, I felt that some of the lectures started from very basic concepts, but then they quickly escalated to much more complex ones, specifically from Module 4. I think there should be a warning before starting this course to ensure people know some Python - if there is a specific course to covers these concepts, it should potentially be part of the certification.

Even for an introductory course most lessons lacked depth. Usually the broad idea of an algorithm is introduced and then an exercise shows a python call to which applies it. However neither are there any theoretical/mathematical insights why the algorithm works, nor does one obtain relevant practical knowledge. E.g. the course fails to even superficially explain the many options and parameters each algorithm has and which are necessary to actually apply it in practice.

What makes it worse is that there is apparently no support and maintenance for this course: There are tons of smaller and some larger mistakes in the lectures as well as the exercises, however reports of those as well as most other questions in the discussion forums remain unanswered.

This course was taught nowhere near as well as the other courses in this certificate track. The code syntax was not explained well at all and it took forever to decipher. The lectures were also not very informative. I would have appreciated a much more in-depth look at the concepts or at least explaining them in further detail. These courses are supposedly for "beginners" but there is no way a "beginner" would be able to get through a course like this without explaining everything better.

The final assignment was also VERY confusing. I would recommend the instructors revisit and revise the course material to make it more engaging and do a better job of explaining the concepts.

Course subject and materials are good, relevant and deep enough. However, this course, as some others in the IBM Data Science track, holds your hand through so much then just drops you on final projects. The final project for this course did not have full enough instructions; the final bit had not been covered at all in earlier weeks and students are left with a generic instruction of 'you should be able to do x'...without any further guidance.

The grammar and English used in the course materials is poor. This makes some learning and assignments unnecessarily difficult, and it's not fair on quizzes/finals to have a question that doesn't make sense in English.

- Could be heavier on the mathematics, which would generate a deeper understanding. - Also if you will not use a specific information, you should not mention it or provide further reading. - Quizes need a rework: answers in some cases are more like "guess what i want to hear", particularly when more than options are definitely correct. - Too often: "Those topics are not in the scope of the course." - The hands on labs are alright all in all another disappointing course...

This course material was good but I think it has some issues:

1- The coding levels in labs are so high and not suitable for beginners.

2- the final exam was simple but it had two issues. The instructor pre-split dataset to train and test parts is confusing without a good explanation and the worst part was the peer-graded section. just provide a reference notebook with confusing rubric grading and had a mistake.

I liked the videos, but there are a lot of mistakes in the notebooks, especially in the solution for the final assignment (which results in unfair gradings). Most of them were mentioned in the forums months ago, but as with all IBM courses, that I have finished so far, no employee seems to care. None of the mistakes gets corrected, and most of the time, you don't even get a reply from one of the moderators.

It could have been very good. But again, one more useless course by IBM. Your task is to copy-paste without asking any question why and how. Graded assignment is a joke. Sample result notebook is useless as nothing is explained, proposed models are bad and NOT CORRECT in a first place. Just give your money to IBM and don't ask questions

the work was insightful, but the peer review of the final project is not working. They don't even look at the work and fail you for no reason.

I'm finishing this certificate program because it would be easier than starting another one from scratch. I've been disappointed with most of the courses and this is no exception. There are mistakes, typos, and poor grammar throughout the course. They have a system to report mistakes, but I should be getting paid to fix your course - not paying to fix it, right? The quizzes are an unnecessary waste of time (they ask very minute, arbitrary questions about videos that are just meant to give you a brief overview).

The labs are the most / only useful aspect of the course because that's where you learn actual code - but they don't explain HOW the code WORKS. They just say what the code does and then they show it to you. There's a difference, as any good teacher knows. This course was clearly created by data scientists, not teachers (and certainly not masters of the English language). I would recommend this certificate program if you already know python and data science and you are just trying to earn a badge that will look good on your resume.

The course is very shallow. It never goes in depth with the algorythms, neither in a mathematical sense, nor in how they are implemented and best used. They don't even cover hyper-parameter optimization using cross-validationThis is not ok for a final course in the IDM Data Science Certificate, especially because Regression was already much better covered in the Data Analysis with Python Course. Moreoever, the Final Assignment features an unbalanced dataset, for which the course does not prepare students enough. If one tries to copy the methods used during the course without reasearching much about this on their own, they will train models that would be unacceptable in a real-world scenario. Worse still, the "model" answer provided does exactly that.